B04: Data analysis challenges for the GRACE Follow-On community

Gravity field recovery missions such as GRACE, typically provide multiple data streams which need to be processed to extract the maximum information.

In GRACE, for example, five major signal streams

  • GPS ranging,
  • K-band ranging,
  • accelerometer data,
  • star camera data and
  • GPS occultation

measurements are recorded for use in the data analysis. Analyzing such complex data sets to extract the full science information requires development and testing of sophisticated algorithms.

So-called “Mock data challenges” (MDCs) are an ideal platform under which to develop and test these algorithms under as realistic conditions as possible, especially before launch where no real data exists. By bringing together system and signal models from other members of the community, data streams can be produced which mimic the behavior of the entire system at a level which allows for rigorous testing of the data analysis algorithms under development. After producing simulated data streams, the entire GRACE community can analyze the data using their various algorithms. Analysis of the results can be used to compare the different algorithms and to help steer the development of future algorithms.

It turns out that, even now, more than 10 years after launch, improvements are still being found and new algorithms are being developed to extract the maximum science from these data streams. In GRACE Follow-On, there will be even more data streams, namely those from the joint U.S.-German Laser Ranging Instrument (LRI) that provides better ranging and pointing measurements, which add to the existing signals from the K-band µ-wave ranging, GPS receivers, clock information from the ultra-stable oscillators, the accelerometers and pointing information from several sensors.

Up to now, no such effort has been carried out in the GRACE community, resulting in groups working independently to solve similar problems. Through MDCs, the sharing of information and core algorithms can be improved, and broader working relationships can be created. Such challenges have already shown themselves to be extremely useful in the LISA community. There, a series of Mock LISA Data Challenges (MLDCs) was run over a number of years (see Arnaud et al. (2007a), Arnaud et al. (2007b), Babak et al. (2008), Babak et al. (2010)). Numerous institutions took part in these challenges and many working groups were formed as a result. This project planned several runs of Mock Data Challenges for the GRACE-FO and future satellite missions which were provided to the geodesy community. The goal was to initiate a basis for more data and methodology exchanges as well as searching answers to the existing questions in the field.

The simulated data for different challenges can be downloaded here: Mock Data Challenge.

The deadline for basic challenge 1 is over and we have collected the solutions from different participants. A paper on the results and comparison is being prepared and scheduled to be submitted by end of April.

The basic challenge 2 contains the LRI data and realistic noise  provided by GRACE FO LRI sensor developers. The deadline for this challenge is 30th June 2017.

The advanced challenges 1 and 2 with more sophisticated simulation scenarios and full LRI noise model are now published and can be used.

Principal Investigators

Dr. Martin Hewitson
Principal Investigator
Address
Max-Planck-Institut für Gravitationsphysik
Callinstr. 38
30167 Hannover
Dr. Martin Hewitson
Principal Investigator
Address
Max-Planck-Institut für Gravitationsphysik
Callinstr. 38
30167 Hannover
Dr. Meike List
Principal Investigator
Address
ZARM
Zentrum für angewandte Raumfahrttechnologie und Mikrogravitation
Dr. Meike List
Principal Investigator
Address
ZARM
Zentrum für angewandte Raumfahrttechnologie und Mikrogravitation